LigSeeSVM: Ligand-Based Virtual Screening Using Support Vector Machines and Information Fusion

摘要

Structure-based virtual screening of molecular compound libraries is a potentially powerful and inexpensive method for the discovery of novel lead compounds for drug development. That said, virtual screening is heavily dependent on detailed understanding of the tertiary or quaternary structure of the protein target of interest, including knowledge of the relevant binding pocket. Such information is often difficult or impossible to obtain. In contrast, ligand-based in silico drug screening can perform screening on drug targets whose three-dimensional structure is not yet determined. We previously demonstrated that the use of data fusion techniques strengthen virtual screening. Building on this data, we here describe LigSeeSVM, a ligand-based screening tool using data fusion and Support vector machines and termed. We combine atom pair (AP) structure descriptors and physicochemical (PC) descriptors to characterize compounds’ features. We used SVM to generate SVM-AP model based on 825 AP descriptors and SVM-PC model based on 19 physicochemical descriptors. We combine SVM-AP and SVM-PC using rank-based information fusion to create LigSeeSVM model. LigSeeSVM was evaluated on five data sets, including thymindine kinase (TK) substrates, estrogen receptor (ER) antagonists, estrogen receptor agonists (ERA), GPCR and GABAA ligands. Our results suggest that LigSeeSVM is useful for ligand-based virtual screening and offers competitive performance to other ligand-based screening approaches.